Bryan Boreham
Bryan Boreham
I have proposed an alternative fix in #13997, which I think is a more general approach.
@ptodev you added this code; what do you think?
Ref: * #11306
If you can supply a heap profile (`curl .../debug/heap` and post the output here), we might be able to see what happened. It's likely that newer versions will show a...
It's not a memory leak; the data structures are just not optimised for this usage. However you should try a newer version; there were a lot more optimisations after 2.45....
> Prometheus version v2.45+stringlabels This is not a tag we published on Dockerhub, and your profile does not match 2.45. It is a much better match to 2.43. From the...
I note that #13147 changes only the experimental service discovery manager, not the one used here.
I can't tell much from that information; if you upload another heap pprof it may give some clues. The way I use the term, a "leak" would show an increase...
That profile shows 43GiB allocated while rendering a web page showing dropped targets. There have been a couple of improvements since v2.43 that should help: * v2.47 added `keep_dropped_targets` which...